mlGCluster: get the gene features by random forest given cluster
In CBIIT-CGBB/GCluster: Graph Based Cluster

mlGCluster

R Documentation

get the gene features by random forest given cluster

Description

Gene feature selection for clustered by random forest given cluster

Usage

  mlGCluster(dat=dat, method=c("randomForest", "xgboost", "party", "CORElearn"), 
                       filter=NULL, cluster=NULL, top.number=200, mtry=2, ntree=50,
                       estimator="ReliefFexpRank", ReliefIterations=100,
                       objective="multi:softprob", eval_metric="mlogloss",    
                       nthread=8, max_depth=16, eta=0.3, gam=0, subsample=1,   
                       colsample_bytree=1, min_child_weight=12, nrounds=20)

Arguments

`dat`	data.frame or matrix, row is by gene and column is for single cell
`method`	any one or more method(s) from packages (eg. randomForest, xgboost, party CORElearn). The methods is(are) named as the package name(s).
`filter`	numeric, top number of most variance genes.
`cluster`	vector, cluster number
`top.number`	numeric, top feature number
`mtry`	party package: cforest_unbiased(mtry)
`ntree`	party package: cforest(cforest_unbiased(ntree))
`estimator`	CORElearn package: attr(estimator)
`ReliefIterations`	CORElearn package: attr(ReliefIterations)
`objective`	xgboost package: xgboost(objective)
`eval_metric`	xgboost package: xgboost(eval_metric)
`nthread`	xgboost package: xgboost(nthread)
`max_depth`	xgboost package: xgboost(max_depth)
`eta`	xgboost package: xgboost(eta)
`gam`	xgboost package: xgboost(gamma)
`subsample`	xgboost package: xgboost(subsample)
`colsample_bytree`	xgboost package: xgboost(colsample_bytree)
`min_child_weight`	xgboost package: xgboost(min_child_weight)
`nrounds`	xgboost package: xgboost(nrounds)